Lead ML/Perception Engineer, May Mobility

• Proven ability to architect and evolve large-scale perception or autonomy systems, balancing performance, latency, safety, and long-term maintainability
• Strong grasp of perception robustness and reliability engineering, including degraded-sensor operation, adverse-weather handling, fault detection, and graceful degradation
• Experience with scene/scenario understanding and runtime activation logic (e.g., context classification, ODD monitoring, behavior triggering)
• Familiar with ML development cycle, deployment, and optimization
• Deep understanding of data: data pipeline, data balancing, data mining, and data-driven performance improvement
• Knowledge of multimodal learning techniques, including contrastive learning and prompt engineering for zero-shot visual recognition
• Deep understanding of testing frameworks and workflows
• Excellent attention to detail and rigorous testing methodology
• Exceptional written and verbal communication skills and team-leading abilities
• A minimum of 5+ years of industry experience working on real-world robot systems maintaining high-quality industrial-grade code
• Master's degree in Robotics, Computer Science, or Computer Engineering, or a field that requires a strong mathematical and/or engineering foundation
• Demonstrated experience leading or making major architectural decisions for a perception, robotics, or autonomy software stack
• Strong programming skills in C/C++/Python; software development in Linux environments
• Strong experience in ML/DL development with PyTorch/TensorFlow
• Direct experience developing or fine-tuning large-scale multi-modal models (e.g., multi-sensor fusion models, foundation models, including VLM backbones) for real-world perception applications
• Experience working on a combination of several of the following real-time areas:
• Computer Vision: object detection, classification, segmentation
• Semantic scene understanding, scene/scenario detection, and open-vocabulary detection
• Perception robustness in degraded or adverse-weather conditions
• Multi-target tracking and sensor fusion
• Extensive experience deploying features/ML/DL models in real-time systems with high accuracy and low latency
• If you’re excited about this role but your previous experience doesn’t align perfectly with every qualification, we encourage you to apply anyway! You may be the perfect candidate for this or another role at May

Desirable

• As Perception Lead, you will own the technical direction of how our vehicles see and interpret the world
• This is a hands-on role focused on three priorities: advancing scene detection so the system reliably recognizes its operating context and engages the right behavior; hardening perception against sensor degradation and adverse weather (rain, snow, fog, glare, low light, and sensor soiling or occlusion) so performance degrades gracefully and safely; and driving major architectural updates to evolve our perception stack into a scalable, maintainable, production-grade foundation
• You will set the technical bar, mentor engineers, and deliver capabilities that operate safely in the real world
• Work independently with cross-functional teams to develop software and system requirements
• Lead major architectural updates to the perception stack, including system-level design, modularization, and migration strategies that improve scalability, latency, and maintainability across the fleet
• Design, develop, and own scene detection and activation capabilities, classifying the operating scene/scenario and triggering the appropriate perception behaviors and operational design domain (ODD) logic at runtime
• Improve perception robustness under degradation and adverse weather, including sensor fault detection, health monitoring, graceful degradation, and fallback strategies for rain, snow, fog, glare, low-light, and sensor soiling/occlusion conditions
• Design, implement, and test state-of-the-art perception features on time with high quality, industrial-grade production code
• Integrate large-scale multi-modal models (including but not limited to VLMs and LLMs) into the perception stack to improve semantic scene understanding and reasoning
• Track and trend technical performance of perception in the field
• Lead major feature development including feature design, code reviews, issue diagnosis, and resolution
• Lead extensive testing to validate features and satisfy release schedules
• Lead development related to data, development, and ML pipelines, specifically focused on multimodal data alignment for training foundation models